ARCHIVES
VOL. 10, ISSUE 1 (2026)
Real-time machine learning model for rheology and viscoelastic behavior under dynamic pressure variation
Authors
Ichenwo John Lander, Ogwu Philip
Abstract
Measured forecast of the rheological and
viscoelastic behavior under dynamic pressure changes will be important in
enhancing the efficiency of the processes and also in avoiding operational
problems in the drilling, polymer processing and pipeline systems. This paper
set out to create a real-time machine learning (ML) model to predict viscosity,
shear stress, yield point and viscoelastic behavior at different pressure and
temperature levels. Measurements in high pressure rheometer were measured at
pressures: 0-10,000 psi and temperatures: 25-120oC and pre-processed
by outlier removal and normalization and feature engineering (including
pressure derivatives and shear history). The hybrid type of model that is a
combination of gradient boosting regression (GBR) and long short-term memory
(LSTM) networks were trained and validated. The outcome is that the hybrid
model had an R2 value of 0.97 and a RMSE of 3.14 in the prediction of
rheological parameters, whereas the traditional Herschel-Bulkley model had an
R2 value of 0.78. The model similarly estimated viscoelastic modulus with R2
= 0.95 of storage modulus (G) and R2 = 0.93 storage modulus (G”).
On-line measurements indicated a mean estimation latency of 78ms, which means
that the system can be used in the sub-second control mode, and thus there is
great prospect of applying it in industries and making upstream decisions
Download
Pages:15-18
How to cite this article:
Ichenwo John Lander, Ogwu Philip "Real-time machine learning model for rheology and viscoelastic behavior under dynamic pressure variation". International Journal of Advanced Engineering and Technology, Vol 10, Issue 1, 2026, Pages 15-18
Download Author Certificate
Please enter the email address corresponding to this article submission to download your certificate.